How hot is the SF housing market, really?

We asked our AI home buying agent, Owen, to analyze every transaction in San Francisco over the last 3.5 years. Owen mapped who is buying what, where the money is coming from, and how competitive each sale is. Then we turned all of that into empirical pricing models. Use Owen to win your next transaction.

San Francisco transactions, 2023 to 2026

8,351San Francisco transactions, 2023 to 2026

buyers matched to an employer and industry

5,344buyers matched to an employer and industry

of Marina buyers come from VC, PE or finance, and it sells below value

24%of Marina buyers come from VC, PE or finance, and it sells below value

of homes are bought by tech employees

40%of homes are bought by tech employees

Tech, AI, VC and PE money is buying 50%+ of homes in SF.

We classified 5,025 employers across 5,344 buyers. Tech has held just under 40% of identified buyers for three years, steady rather than surging. The genuine movement is underneath it: capital is rotating from operators to investors. And the AI money has not come in yet.

Venture capital and private equity is the fastest-growing buyer group in the city, up 1.41 points to 3.8% of buyers. Financial services adds 0.96. Artificial intelligence adds 0.84 to reach 6.6%. On the other side, the operating tech that built the city is receding: internet and e-commerce down 1.32 points, software and IT down 1.27.

San Francisco is not being bought by more tech people. It is being bought by the people who fund them.

Every neighborhood has a buyer profile. They are nothing alike.

Aggregate shares hide the real structure. At neighborhood level the concentration is extreme, and it is stable enough to price against.

Marina draws 24% of its buyers from venture capital, private equity and finance, nearly two and a half times the citywide rate. Mission Bay is 53.8% tech. AI buyers cluster somewhere else again: Twin Peaks, Potrero Hill and Inner Sunset, all well above the 6.2% citywide share.

Which raises the obvious question. If you know where the money is, do you know where the competition is?

Buying power varies widely.

But the most competitive transactions are at the lower end of the pricing range.

VC and PE buyers pay the most by a distance, a $2.12M median, 29% above the all-buyer figure and $475K clear of the next group. They are also the least likely to be buying the competitive end of the market: their median Owen Index is 39.9, near the bottom of the table. They buy expensive, low-contention property in the wealth corridor.

The inversion is at the other end. Government Administration buyers post the highest median Owen Index in the dataset (66.0) on the lowest median price ($1.25M). Primary and Secondary Education is second (63.1 on $1.31M). Public sector households are concentrated in precisely the ZIPs where bidding is fiercest relative to valuation: the Excelsior, Visitacion Valley, the outer Sunset. They compete hardest for the cheapest homes, while fund principals compete least for the dearest.

VC and PE clusters in the Marina and Pacific Heights (94123, 94115). Government workers concentrate in the Excelsior, 16.4% of them in a single ZIP. AI buyers look like the rest of tech: the Mission, the Castro, and the affordable west.

Cash does not buy a discount. It pays a premium.

Everyone in San Francisco believes cash wins. It does not win on price. Across 8,328 sales, adjusted for neighborhood, property type, condition, size, price tier and quarter, cash buyers paid 0.95 points more relative to fair market value than financed buyers, roughly $14,700 on a median sale. Cash here is not a negotiating instrument. It is what wealthy and institutional buyers use to compete for scarce property, and they pay up to do it. The caveat on this analysis: since we don't have data on lost offers, we can't estimate the odds that a cash bid beats a financed one at the same price.

The only place cash buys a discount is the one place financing might genuinely fall through: below average-condition homes, where it takes 2.2 points off, about $34,700.

Every listing takes a position. Most take the same one.

San Francisco Realtors generally follow the same strategy: underprice the listing by 5 to 15% relative to Fair Market Value to invite a bidding war. Does it work?

Of the sellers who asked below fair market value, 40% cleared it. Of those who asked above, 79% did.

Among agents with five or more verified sales, the underpricing habit is strikingly persistent. Owen allows you to model the pricing strategy of every listing agent in the city and optimize how to bid.

A pricing agent for any San Francisco address.

Everything above is what the market looks like in aggregate. Owen answers it for the one address you care about, before it is listed, in the time it takes to type it.

Owen scores an address on what is knowable before listing: neighborhood, property type, condition, age, size, price tier and commute geometry. Market timing, pricing strategy and the agent are deliberately excluded from the score, because those are the things you control. It is a pricing analysis, not an appraisal.

About the data. 8,351 San Francisco residential transactions recorded January 2023 to August 2026. Pricing findings rest on the 1,931 with an asking price verified against the record: the MLS overwrites list price with sale price when a listing closes, so we reconstructed the original ask and validated it at 96.4% exact match. Buyer analysis covers 5,344 buyers matched to 5,025 classified employers. Figures are aggregated to neighborhood and quarter; no individual buyer, seller, agent or address is identified. Owen produces pricing analysis, not an appraisal, and is not a substitute for professional valuation or legal advice.